Adam Wenchel, Arthur.ai | CUBE Conversation
summary
Company builds a production monitoring platform that validates and guards LLMs and foundation models before and after deployment, flags hallucinations, bias, data drift and upstream-data breaks, and ties model metrics to business KPIs. Origin story traces from DARPA work and a Capital One AI team that handled hundreds of millions in revenue. Integrates with SageMaker and common ML stacks, reduces manual checking that consumes 25–30% of data scientist time, and runs a DSOC (Data Science Operations Center) for enterprises including top 10 US banks and Fortune 100 healthcare and retail customers.
AI-written summary · about
